Application of Multivariate Non-parametric Change-point Control Charts to Children with Bronchial Pneumonia

Abstract

Control charts are statistically and visually designed to detect changes or shifts in process. We apply the combined two nonparametric Rank Test of Wilcoxon-Mann-Whitney and Mood statistics called Lapage-type Change-Point (LCP) Chart. The modified chart shows great efficiency in detecting signals and shift in children’sbronchial pneumonia, at about two years old (24th months), while it suggests that the actual shift had started at the 21st month (observation). This is an indication of the LCP promptness in raising alarm of a process shift if indeed it exists. The signal may have resulted due to shift in both mean (p=0.002949) and variability (p=0.03978) of children’s bronchial pneumonia as measured. The study suggests that the new method should be used in short-run situations since it has the capacity of not only detecting shift but also the period it occurs and also where the underlying distributions are usually unknown.